Unravelling Abiogenic and Biogenic Sources of Methane in the Earth's

Chemical Geology 226 (2006) 328–339 www.elsevier.com/locate/chemgeo

Unravelling abiogenic and biogenic sources of methane in the Earth’s deep subsurface

B. Sherwood Lollar a,*, G. Lacrampe-Couloume a,1, G.F. Slater a,1, J. Ward a,1, D.P. Moser b,2, T.M. Gihring b,3, L.-H. Lin c,4, T.C. Onstott c,4

a Department of Geology, 22 Russell St., University of Toronto, Toronto, Ontario, Canada M5S 3B1 b Environmental Microbiology Group, Pacific Northwest National Laboratory, Richland WA 99352, USA c Dept. of Geosciences, Guyot Hall, Princeton University, Princeton NJ 08544, USA Accepted 5 September 2005

Abstract

At four underground sites in Precambrian Shield rocks in Canada and South Africa, hydrocarbon and hydrogen gases exsolving from saline fracture waters are analyzed for compositional and isotopic signatures. Dominated by reduced gases such as CH4,H2 and higher hydrocarbons (ethane, propane, butane), the most 13C-enriched methane end-members at all four sites show a pattern of carbon and hydrogen isotopic values similar to abiogenic gases produced by water–rock interaction that have been identified previously at one site on the Precambrian Shield in Canada. The abiogenic nature of these gases was not previously recognized due to mixing with a second methane component produced by microbial processes. The microbial methane end-member is identified based on carbon and hydrogen isotopic signatures, and DNA gene amplification (PCR) data that indicate the presence of methanogens. A framework is presented to estimate the relative contribution of abiogenic versus microbial hydrocarbon gases at these sites. This approach has important implications for evaluation of potential abiogenic hydrocarbon reservoirs in a wide range of geologic settings, including the longstanding controversy concerning the possible contribution of abiogenic gases to 13 economic petroleum hydrocarbon reservoirs. The association of high concentrations of H2 with C-enriched CH4 end-members, 13 and H2 depletion in the C-depleted methanogenic end-members further suggests the possibility that abiogenic gases may support H2 autotrophy linked to methanogenesis in the deep subsurface. D 2005 Elsevier B.V. All rights reserved.

Keywords: Abiogenic; Methane; Hydrogen; Autotrophy; Deep biosphere; Mars

* Corresponding author. Tel.: +1 416 978 0770; fax: +1 416 978 3938. E-mail addresses: [email protected] (B. Sherwood Lollar), [email protected] (G. Lacrampe-Couloume), [email protected] (G.F. Slater), [email protected] (J. Ward), [email protected] (D.P. Moser), [email protected] (T.M. Gihring), [email protected] (L.-H. Lin), [email protected] (T.C. Onstott). 1 Fax: +1 416 978 3938. 2 Fax: +1 702 862 5360. 3 Fax: +1 850 644 2581. 4 Fax: +1 609 258 1274.

1. Introduction workings at rates of 1 to N30 L gas/min/borehole (Sherwood Lollar et al., 1993a,b). The groundwaters Recent reports of CH4 in the Mars atmosphere fo- are typically NaCaCl-rich or CaNaCl-rich with low cussed scientific and public attention on possible geo- levels of SO4 (typically b15 ppm and always b100 logical and biological sources of these hydrocarbons ppm) with pH levels between 7–9. This study focuses (Formisano et al., 2004; Kerr, 2004). Resolving the on two sites in Canada (Kidd Creek and Copper Cliff question of the origin of atmospheric CH4 on Mars is South mines) and three in South Africa (Driefontein, even more challenging given that distinguishing abio- Kloof and Mponeng mines). Kidd Creek mine, situated genic versus biogenic sources of CH4 in the terrestrial in the southern volcanic zone of the Abitibi greenstone subsurface is still controversial (Gold, 1979; Kenney et belt (approximately 2700 Ma), 24 km N of Timmins, al., 2002; Shock, 1995). Deep subsurface fluids in Ontario is one of the world’s largest volcanogenic Precambrian Shield rocks have been shown to be dom- massive sulfide deposits. All samples were collected inated by reduced gases such as CH4 and locally, high from 2–2.1 kmbls (kilometers below land surface). The concentrations of H2 (up to 30% by volume) (Sherwood Cu–Ni ore at Copper Cliff South mine (CCS) in Sud- Lollar et al., 1993a,b). Sherwood Lollar et al. (2002) bury, Ontario is associated with a quartz diorite dike used stable isotope signatures to suggest that CH4 and offset of the Sudbury Igneous Complex — a micro- higher hydrocarbon gases (ethane, propane and butane) pegmatite, norite and quartz diorite irruptive emplaced at Kidd Creek mine on the Canadian Shield are pro- approximately 1840 Ma (Cochrane, 1984) and samples duced abiogenically by water–rock interaction such as were from approximately 1.3 kmbls. In South Africa, surface-catalysed polymerization (Anderson, 1984; Kloof, Driefontein and Mponeng mines are all located Foustoukos and Seyfried, 2004); metamorphism of west of Johannesburg in the Witwatersrand Basin — a graphite–carbonate bearing rocks (Giardini and Salotti, large Archean intracratonic basin composed of volca- 1969; Holloway, 1984; Kenney et al., 2002); and other nosedimentary sequences that unconformably overlie gas–water–rock alteration reactions such as serpentini- 3.0 Ga granitic basement. The basin is divided chrono- zation (Berndt et al., 1996; Charlou and Donval, 1993; logically into quartzite, shale and minor volcanic strata Horita and Berndt, 1999; Kelley et al., 2001, 2005; of the Witwatersrand Supergroup (2900 Ma), the basal- McCollom and Seewald, 2001; Vanko and Stakes, tic lava sequence of the Ventersdorp Supergroup (2700 1991). In this paper, data from 4 new sites in Canada Ma) and the overlying clastic and dolomitic sediments and South Africa suggests that abiogenic hydrocarbon of the Transvaal Supergroup (2400–2500 Ma) (Coward gases are more globally pervasive than has been under- et al., 1995). Almost all the samples in this study are stood previously. This is due to the fact that at many from the Ventersdorp Supergroup and are located be- sites, the distinct abiogenic signature of such hydrocar- tween 2.7 and 3.4 kmbls. The exceptions are borehole bons is obscured by mixing with microbial CH4. Based DR938H3, which while drilled in the Ventersdorp, on this data, we present a model for identification of intersects the underlying Witwatersrand Supergroup at abiogenic hydrocarbons through the resolution of mi- approximately half its 750 m length, and DR9IPC, crobial and abiogenic mixing — an approach that is which is drilled in the overlying Transvaal Supergroup applicable to a wide variety of crustal settings where and is located at 0.9 kmbls. potential abiogenic hydrocarbon reserves have been suggested (Gold, 1979; Kenney et al., 2002; Shock, 3. Methodology 1995). 3.1. Sampling methods 2. Geological setting and samples All gas and fracture water samples were collected at In gold and base-metal mines throughout the Pre- the borehole collar after the method of Sherwood Lollar cambrian Shield rocks of Canada, Finland and South et al. (2002) and Ward et al. (2004). A packer was Africa, flammable gases discharge from fractures and placed into the opening of the borehole and sealed to exploration boreholes (Nurmi and Kukkonen, 1986; the inner rock walls below water level to seal the Lahermo and Lampen, 1987; Cook, 1998; Sherwood borehole from the mine air and minimize air contami- Lollar et al., 1993a,b). Originally dissolved in saline nation. Gas and water were allowed to flow through the groundwater (TDS levels range from several 1000 to apparatus long enough to displace any air remaining in tens of 1000s of ppm) in sealed fracture systems in the the borehole or the apparatus before sampling. Plastic rocks, gases are released via depressurization into mine tubing was attached to the end of the packer and the 330 B. Sherwood Lollar et al. / Chemical Geology 226 (2006) 328–339

flow of gas and/or water from the borehole was directed concentrations of the inorganic gas components (H2, into an inverted graduated funnel. Gases collected in He, Ar, O2,CO2 and N2). To determine concentrations the inverted funnel were transferred directly into evac- of Ar, O2 and N2 the helium gas flow rate was 3 ml/min uated vials through a needle that was attached to the top and the temperature program was: initial 30 8C hold 6 of the funnel. The gas sampling vials were pre-evacu- min, increase to 80 8Cat158C/min, hold 4 min. To ated 130 ml borosilicate vials sealed with butyl blue determine CO2 concentrations, the helium gas flow rate rubber stoppers prepared after the method of Oremland was 50 ml/min and temperature program was: initial 60 and Des Marais (1983). Vials were pre-fixed with 50 8C, increase to 250 8Cat208C/min, hold 6 min. To Al of a saturated HgCl2 solution to kill any microbes determine concentrations of H2 and He, the argon contained in the sample so microbial activity post-sam- carrier gas flow rate was 2 ml/min and temperature pling would not alter the gas composition and isotopic program was: initial 10 8C hold 10 min, increase to signatures. Previous studies, comparing the isotopic 80 8Cat258C/min, hold 7 min. All analyses were run values of gases taken at the borehole collars to values in triplicate and mean values are reported in Table 1. determined for gases in solution at depth in the same Reproducibility for triplicate analyses was F5%. boreholes, showed that exsolution of the hydrocarbon gases from solution does not alter their isotopic signa- 3.3. Isotopic analysis tures (Sherwood Lollar et al., 1993a,b, 1994). For the South African sites, the packer was sterilized Stable carbon and hydrogen isotopic analysis for all by autoclaving before insertion into the borehole in hydrocarbons were performed at the University of Tor- order to reduce the possibility of introducing surface onto (Table 2). Analyses for d13C values were per- microrganisms by that means. In mining environments, formed by continuous flow compound specific carbon aseptic drilling techniques were not possible. Microbial isotope ratio mass spectrometry with a Finnigan MAT populations were characterized and compared however 252 mass spectrometer interfaced with a Varian 3400 between the saline fracture waters from these sites, the capillary GC. Hydrocarbons were separated by a Por- low salinity mine service water, and ventilation air. aplot Qk column (25 m0.32 mm ID) with temper- Comparison of 16S rRNA gene clone libraries from ature program: initial 40 8C hold 1 min, increase to 190 these possible sources of contamination and from the 8Cat58C/min, hold 5 min. To separate CO2 from CH4 fracture waters yielded no commonalities (Kieft et al., the program was started at 10 8C (hold 2 min) increased 1999; Onstott et al., 2003; Takai et al., 2001). In to 190 8Cat58C/min (hold 5 min). Total error incor- addition the fact that dissimilatory sulphate reductase porating both accuracy and reproducibility is F0.5x (DSR) and 16S ribosomal (rRNA) genes amplified with respect to V-PDB standard. from the fracture water DNA yield sequences that are The d2H analysis was performed on a continuous most similar to those reported for sulphate-reducing flow compound specific hydrogen isotope mass spec- bacteria isolated from other subsurface sites is consis- trometer which consists of an HP 6890 gas chromato- tent with the fracture water microbial populations being graph (GC) interfaced with a micropyrolysis furnace in situ subsurface communities and not surface con- (1465 8C) in line with a Finnigan MAT Delta+-XL taminants (Baker et al., 2003). isotope ratio mass spectrometer. The hydrocarbon gases were separated on a Poraplot Qk column (25 3.2. Compositional gas analysis m0.32 mm ID) with a helium carrier at 2.2 ml/min and temperature program: initial 35 8C hold 3 min, Compositional analyses of gas samples were per- increase to 180 8Cat158C/min. Total error incorpo- formed at the Stable Isotope Laboratory at the Univer- rating both accuracy and reproducibility is F5x with sity of Toronto. A Varian 3400 GC equipped with a respect to V-SMOW. flame ionization detector (FID) was used to determine concentrations of CH4,C2H6,C3H8 and C4H10. The 3.4. DNA gene amplification (PCR) hydrocarbons were separated on a J and W Scientific GS-Q column (30 m0.32 mm ID) with a helium gas Methyl coenzyme M reductase a-subunit genes flow and temperature program: initial 60 8C hold 2.5 (mcrA) were amplified from community DNA extracts min, increase to 120 8Cat58C/min. A Varian 3800 GC according to the methods of Hales et al. (1996).Ar- equipped with a micro-thermal conductivity detector chaeal 16S rRNA gene was amplified using a nested (ATCD) and a Varian Molecular Sieve 5A PLOT col- PCR reaction in which PCR product generated using umn (25 m0.53 mm ID) was used to determine the primers 21F (Moyer et al., 1998), and 1492R (Rey- B. Sherwood Lollar et al. / Chemical Geology 226 (2006) 328–339 331

Table 1 Gas compositional data (in %)

Site Borehole No. Ar H2 He O2 N2 CH4 C2H6 C3H8 C4H10 Copper Cliff CCS4577 0.06 54.0 3.46 0.11 1.92 34.4 5.31 0.49 0.11 Copper Cliff CCS4547 0.07 43.0 3.38 0.12 2.20 38.1 5.28 0.56 0.12 Copper Cliff CCS4546 0.17 9.94 4.37 0.52 6.38 69.5 7.14 0.78 0.17 Copper Cliff CCS4572 0.05 57.8 2.62 0.15 1.72 25.3 4.16 0.38 0.09 Copper Cliff CCS4880 0.27 19.7 6.42 0.10 4.54 59.3 7.39 1.11 0.29 Mponeng MPA NA 3.30 9.06 4.77 28.1 53.8 2.47 0.36 0.08 Mponeng MP104 NA 11.5 12.3 3.65 21.1 49.6 3.99 0.60 0.13 Driefontein DR548 NA 10.3 3.05 6.18 23.6 50.7 3.85 0.52 0.11 Driefontein DR938H1 NA b0.01 0.90 16.5 72.5 5.45 0.14 0.02 b0.01 Driefontein DR938H3-0m NA 0.74 5.98 0.55 14.0 76.0 3.15 0.32 0.06 Driefontein DR938H3V-0m* NA 0.32 4.64 4.95 28.6 61.4 2.45 0.26 0.04 Driefontein DR938H3-648 m NA NA NA NA NA NA NA NA NA Driefontein DR938CH1 NA b0.05 0.21 23.4 89.9 1.19 b0.05 b0.05 b0.05 Driefontein DR9IPC NA b0.05 2.75 3.77 81.9 11.6 b0.05 b0.05 b0.05 Kloof KL1GH NA 9.15 3.42 14.5 56.4 10.7 0.49 b0.01 b0.01 Kloof KL441H3 NA b0.01 15.9 0.43 45.1 33.0 0.66 0.07 0.02 Kloof KL739 NA 9.25 13.5 b0.04 7.72 64.9 2.86 0.41 0.08 Kloof KL441H2 1.68 b0.01 15.6 6.55 61.9 23.4 0.51 0.05 b0.01 Kloof KL443HWND1 1.39 b0.01 18.2 1.24 13.3 57.3 2.03 0.28 0.05 Kloof KL443HWDN 1.79 b0.01 20.8 0.97 16.8 53.4 2.04 0.21 0.03 a. All CO2 was below detection limit. b. NA — not analyzed. c. C4H10 incorporates both n-butane and iso-butane. d. *DR938H3V-0 m is a sample from the same borehole as DR938H3-0 m taken 8 months afterwards. Such temporal sampling was not possible for the other boreholes due to the common practice of sealing them immediately after completion of drilling. e. DR938H3-0 m and DR938H3-648 m were collected from the same borehole on the same day. The 0 m sample was taken at the borehole collar while the 648 m sample was taken at 648 m depth. senbach et al., 2000) was used as template in PCR distinguished from mantle-derived abiogenic gases reactions with the bnestedQ primers 21F and 958R (Jenden et al., 1988; Jenden and Kaplan, 1989; Poreda (Moyer et al., 1998). McrA and archaeal 16S rRNA et al., 1986, 1988; Lyon and Hulston, 1984; Rigby and gene PCR products were cloned using Invitrogen Smith, 1981). Over the past decade however, there has TOPO kits. For each library, 48 clones were screened been a growing body of literature that indicates that not by restriction fragment length polymorphism. Unique all abiogenic gases are mantle-derived. A variety of low clones were sequenced, then analyzed using BLAST temperature water–rock interactions have been shown (Altschul et al., 1997) to identify sequences in the to produce both CH4 and higher hydrocarbons such as NCBI Genbank database having high similarity (Table ethane, propane, and butane — including surface-cata- 3). Sequence identities were calculated using BioEdit. lyzed polymerization from reduction of CO or CO2 in a Fischer–Tropsch synthesis; heating or metamorphism 4. Results and discussion of graphite or carbonate bearing rocks; or other vapor–water–rock alteration reactions such as serpenti- 4.1. Isotopic patterns suggest an abiogenic origin nization (Holloway, 1984; Yuen et al., 1990; Berndt et al., 1996; Hu et al., 1998; Horita and Berndt, 1999; Mantle-derived abiogenic hydrocarbons are typically McCollom and Seewald, 2001; Foustoukos and Sey- 13 identified based on three criteria: a d C value for CH4 fried, 2004). Significantly, several of these studies have more enriched than 25x;abcarbon isotopic reversalQ demonstrated that production of abiogenic CH4 by trend of increasing isotopic depletion in 13C with in- these water–rock interactions can result in d13C values creasing molecular weight for CH4–ethane–propane– as depleted as 57x (Yuen et al., 1990; Hu et al., butane; and a 3He/ 4He ratio indicative of mantle-de- 1998; Horita and Berndt, 1999). In a crustal-dominated rived helium (R /RaN0.1) (Jenden et al., 1993). Based geologic setting, these processes will produce abiogenic on these criteria, coal bed or thermogenic hydrocarbon hydrocarbons whose d13C values reflect local crustal gases with anomalously enriched d13C values and/or carbon sources and will not have either 13C-enriched anomalously depleted d2H values for methane could be d13C values or R /Ra values N0.1. In addition, it has 332 B. Sherwood Lollar et al. / Chemical Geology 226 (2006) 328–339

Table 2

Carbon and hydrogen isotopic values (in x) for CH4–C4 hydrocarbon gases 13 2 13 2 13 2 13 2 Site Borehole no. d CCH4 d HCH4 d CC2 d HC2 d CC3 d HC3 d CnC4 d HnC4 Copper Cliff CCS4577 36.1 413 24.6 266 ND ND ND ND Copper Cliff CCS4547 34.4 424 28.1 290 ND ND ND ND Copper Cliff CCS4546 32.0 452 34.6 348 33.4 ND ND ND Copper Cliff CCS4572 36.7 401 24.4 269 ND ND ND ND Copper Cliff CCS4880 37.3 372 31.6 301 ND ND ND ND Mponeng MPA 27.8 349 35.0 267 33.3 212 ND ND Mponeng MP104 32.8 366 37.6 270 34.8 193 35.3 88 Driefontein DR548 46.5 403 50.5 285 47.2 192 45.6 72 Driefontein DR938H1 53.6 295 42.9 326 42.3 ND ND ND Driefontein DR938H3-0 m 40.2 368 43.1 290 42.0 ND ND ND Driefontein DR938H3V-0 m* 42.3 369 ND ND ND ND ND ND Driefontein DR938H3-648 m 40.9 371 44.0 234 42.9 179 ND ND Dreifontein DR938CH1 47.5 329 43.2 ND ND ND ND ND Driefontein DR9IPC 55.5 218 ND ND ND ND ND ND Kloof KL1GH 29.2 312 33.2 230 ND ND ND ND Kloof KL441H3 40.0 253 30.9 241 25.2 ND ND ND Kloof KL739 28.7 300 30.7 230 27.2 142 ND ND Kloof KL441H2 37.8 257 32.9 198 34.0 105 ND ND Kloof KL443HWND1 34.4 327 35.6 251 33.7 208 ND ND Kloof KL443HWDN 34.3 329 34.3 269 33.5 206 33.6 207 a. ND — below detection limit for isotope analysis. b. *DR938H3V-0m is a sample from the same borehole as DR938H3-0m taken 8 months afterwards. Such temporal sampling was not possible for the other boreholes due to the common practice of sealing them immediately after completion of drilling. c. DR938H3-0m and DR938H3-648m were collected from the same borehole on the same day. The 0 m sample was taken at the borehole collar while the 648 m sample was taken at 648 m depth. become clear that the bcarbon isotopic reversalQ trend into larger hydrocarbon chains in polymerization reac- alone is not sufficient evidence to support an abiogenic tions, whereas, owing to preferential cleavage of the origin (Horita and Berndt, 1999). weaker 12C–1H bond versus the 12C–2H bond, the Based on crustal 3He/ 4He ratios at all sites, no light (1H) isotope will be preferentially eliminated mantle-derived component for the Precambrian Shield (Sherwood Lollar et al., 2002). In contrast, the pattern hydrocarbon gases is indicated (Lippmann et al., 2003; for commercial thermogenic natural gas deposits con- Sherwood Lollar et al., 1993b). Sherwood Lollar et al. sists of a positive correlation of d13C and d2H values (2002) suggested a crustal abiogenic origin due to between CH4, ethane and the higher hydrocarbon gases water–rock interactions for the hydrocarbon gases at propane and butane (Fig. 1A). This isotopic pattern of Kidd Creek however, based on a new criteria — an increasing isotopic enrichment in both 13C and 2H with 13 2 inverse correlation of C-depletion and H-enrichment increasing molecular weight for the C1–C4 homologues between CH4 and ethane (Fig. 1A). Building on a for thermogenic gases results from production of hydro- hypothesis for carbon isotope distribution patterns first carbons by thermal cracking of a high molecular weight proposed by Des Marais et al. (1981), it was suggested organic precursor (Des Marais et al., 1981; Schoell, that this pattern results from synthesis of higher molec- 1988). At each of the 4 newly investigated Precambrian ular weight hydrocarbons from CH4 during polymeriza- Shield sites in Canada and South Africa from this study, 12 13 tion. In such a process CH4 reacts faster than CH4 to hydrocarbon gases were identified with the same pattern 12 13 2 form chains, so that C is more likely to be incorporated of C-depletion and H-enrichment between CH4 and

Table 3 16S rRNA gene and mcrA clones from borehole DR9IPC having high relatedness to known methanogenic organisms DR9IPC clone (Genbank accession number) Percent of library Nearest cultivated relative Identity (%) 16S rRNA 1 (AY604061) 40.0 Methanobacterium curvum 93 mcrA 1 (AY604057) 23.7 Methanosarcina barkeri 99 mcrA 2 (AY604058) 73.7 Methanobacterium bryantii 86 mcrA 3 (AY604059) 2.7 Methanobacterium aarhusense 83 B. Sherwood Lollar et al. / Chemical Geology 226 (2006) 328–339 333

and 2). In all cases, the proposed abiogenic signature was observed only for the samples with the most 13C- enriched CH4 values at each site (DR938H3 at both 0 and 685 m, DR548; KL739 and KL1GH; CCS4546; and both samples from Mponeng — MP104 and MPA; Fig. 1B). The distinct 13C–2H isotopic patterns between CH4 and ethane for these samples (Fig. 1B), as well as their significantly depleted d2H values compared to typical thermogenic gases (Fig. 2), do not support a thermogenic origin for these 13C-enriched end-members. Furthermore, as at Kidd Creek, the majority of the data points from CCS, Driefontein and Kloof Mines fall along linear trends between the most 13C-enriched and 2H-depleted end-members, and samples with more 13C-depleted and 2H-enriched values (Fig. 2), raising the possibility of mixing between these two end-members. For CCS, Kloof and Kidd Creek, all data fall along a single trend from the 13C-enriched end-members identified in Fig. 1B, whereas for Driefontein, two possible trends can be drawn through the data, one for DR938H3 and one for DR548. At Mponeng, both 13 2 13 Fig. 1. Plot of d C versus d H values for C1–C4 for Kidd Creek (A) samples MP104 and MPA are C-enriched and show and for 4 new Precambrian Shield sites (B). In the lower panel, the 13 the proposed abiogenic signature (Fig. 1B) and this is most C-enriched samples at each site — Driefontein (open squares); consistent with the lack of any mixing trend in Fig. 2. Mponeng (closed circles), Kloof (closed triangles) and CCS (crosses) all show the characteristic 13C depletion and 2H enrichment between In Fig. 2, solid lines indicate the conventionally ac- 13 2 CH4 and ethane similar to the proposed abiogenic gases at Kidd Creek cepted fields of d C and d H values for microbial and (data from (Sherwood Lollar et al., 2002) – closed squares– upper thermogenic CH4 based on empirical studies (Schoell, panel). In marked contrast, the pattern for thermogenic gas reservoirs 13 13 2 1988). The hatched trend lines indicate that C-depleted is a positive correlation of d C and d H values for CH4 and ethane and 2H-enriched end-members at these sites could theo- due to increasing isotopic enrichment in both 13C and 2H with increasing molecular weight for the C –C homologues (open circles retically be either thermogenic or microbial in origin 1 4 13 2 — upper panel). This pattern results from production of the hydro- based on the distribution of the d C and d H fields. carbon gases by thermal cracking of a high molecular weight organic Other lines of evidence indicate that this end-member is 13 precursor (Des Marais et al., 1981; Schoell, 1988). Error bars on d C microbial however, and suggest that microbial CH mix- F x 2 F 4 values are 0.5 . Error bars on both A and B on d H values are 5 ing with abiogenic gases produces the observed trend x and are smaller than the plotted symbols. lines. For borehole DR9IPC, 16S rRNA gene cloning and mcrA (methyl coenzyme M reductase) gene cloning ethane as the proposed abiogenic gases from Kidd Creek indicate the presence of methanogens (Table 3). In addi- (Fig. 1B; Table 2). Interestingly, for each site only the tion, the d13C and d2H values for DR9IPC are similar to 13 most C-enriched samples show this pattern. the values for microbial CH4 identified in several other sites in the Witwatersrand Basin (Ward et al., 2004). In 4.2. Mixing trends particular, fracture water from Masimong and Merrie- spruit, located 150 km southwest of Driefontein, have 13 2 Whereas the Kidd Creek data set is remarkably d C and d H values for CH4 of 60.7 and 207x 13 uniform, one sample (KC7792) has a d CCH4 value (MM5); and 53.7 and 194x (MR1), respectively 13 2 significantly more depleted in C and enriched in H (Fig. 2). The microbial origin of CH4 from Merriespruit than other samples from this site (Fig. 1A). Sherwood is also supported by 16S rRNA gene and by enrichment Lollar et al. (2002) suggested that this might be due to cultures which indicate the presence of methanogens mixing with a more 13C-depleted and 2H-enriched mi- (Ward et al., 2004). crobial CH4 component in this borehole. In the current In contrast, 16S rRNA gene amplifications for the study, the hydrocarbon gas samples analysed for 4 new Kloof samples (KL739, KL441H2, KL443HWND1, Precambrian Shield sites show a much wider range of KL443HWDN) did not yield any evidence of methano- CH4 isotopic signatures than at Kidd Creek (Figs. 1B gens (Ward et al., 2004; Kieft et al., 2005). While this 334 B. Sherwood Lollar et al. / Chemical Geology 226 (2006) 328–339

13 2 Fig. 2. Plot of d C versus d H values for CH4 for sites from this study compared to the conventional fields for microbial and thermogenic CH4 after Schoell (1988). Except for the two Mponeng samples (see text), samples fall along linear trends from the most 13C-enriched end-members to more 13C-depleted and 2H-enriched end-members. Symbols are as in Fig. 1. DR938H3 and DR938H3V are samples taken from the same borehole 8 months apart. Unfortunately such temporal sampling was not possible for the other holes, due to the common practice of sealing them immediately after completion of drilling. Error bars are F0.5 x on d13C values and F5 x on d2H values and are smaller than the plotted symbols. Data from Masimong (MM) and Merriespruit (MR) are from Ward et al. (2004) for comparison. Data from Kidd Creek are from Sherwood Lollar et al. (2002).

may be due to the very low biomass in these samples and end-member, while microbial CH4 has been identified at does not entirely rule out a small microbial CH4 contri- Kidd Creek and Copper Cliff South mines based on bution, it is nonetheless consistent with the predominant- microbial cultures (Doig, 1994), no information on the ly abiogenic origin proposed for CH4 in the Kloof isotopic composition of that CH4 exists to date. Whereas samples. No microbiological work was carried out at a microbial end-member has been identified at Driefon- the other boreholes at Kloof, Copper Cliff South or tein (DR9IPC) based on both isotopic evidence and gene Mponeng, but for Driefontein, 16S rRNA gene amplifi- cloning (Table 3), we do not have the same information cation and culture enrichments for methanogens were for Kloof. The same dolomitic Transvaal Supergroup attempted for both borehole DR9IPC (results above) and overlies Kloof mine however and as at Driefontein, for one of the proposed abiogenic end-members, bore- fractures intersecting both the Ventersdorp and overlying hole DR938H3. While enrichments were unsuccessful, Transvaal provide a conduit for mixing of fluids and PCR-amplified 16S rRNA gene indicated a bacterium gases between these two formations. For all samples (closely related to the sulfate-reducing Desulfotomacu- then, as an initial test of the model, we used a d13C lum species) and three archaeal taxa within the family value of 55x for the microbial end-member based Methanobacteriaceae. The low biomass in this sample on the carbon isotope value of DR9IPC (Fig. 3). By suggests that concentrations of CH4 in the 100 uM range this approach, data from Kloof are consistent with mix- derived from microbial methanogenesis would be insuf- ing of the most 13C-enriched end-member (KL739) with ficient to isotopically overprint the N17 mM abiogenic up to 43% of a microbial hydrocarbon end-member with 13 CH4 signature however (Moser et al., in press). a d CCH4 value of 55x (Fig. 3). All estimates are The possibility of mixing between microbial hydro- insensitive to whether the microbial end-member has 13 2 + 3 4 carbons (characterized by C-depleted and H-enriched CH4 /C2 ratios of 10 –10 or even higher. For Driefon- CH4 values) and abiogenic end-members can be further tein, DR938H1 is consistent with mixing with up to 84% evaluated using the traditional scheme shown in Fig. 3, microbial CH4 (based on mixing with DR548), while adapted from Hunt (1996). In this plot, microbial hydro- mixing between DR9IPC and DR938H3 defines a sep- carbons fall in the upper left corner based on their 13C- arate trend line, as in Fig. 2. Given that the gases occur in + depleted CH4 values and CH4 /C2 ratios N1000 (Hunt, hydrogeologically isolated fractured rock it is not sur- 1996). We have modified the scheme to evaluate abio- prising that some sites exhibit a range of end-members genic-microbial mixing, using the abiogenic end-mem- and different possible mixing lines, as fracture controlled bers for each site suggested by Fig. 1B. For the microbial systems are rarely homogeneous. In fact, the mixing B. Sherwood Lollar et al. / Chemical Geology 226 (2006) 328–339 335

what. The key point of Figs. 3 and 4 is not necessarily to constrain a specific microbial d13C value for each site, but to demonstrate that the isotopic and compositional variation in samples observed supports a model of mixing between the 13C-enriched end-members in Fig. 1B, and microbial end-members with a range of d13C values typical of CH4 produced by methanogens. If a similar approach is taken for the Canadian Shield sites, KC7792 can be explained by a mixture of the most 13C-enriched end-member at Kidd Creek (KC8558), with up to 17.5% microbial CH4 (Fig. 3). The CCS data fit somewhat less well. CCS4546 is the only sample with abiogenic char- acteristics at this site (Fig. 1B) and is the most 13C- enriched and 2H-depleted value at the site (Fig. 2). + Hence, whereas the CH4 /C2 ratio for CCS4546 in Fig.

13 + 3 is slightly elevated (8.6 versus approximately 6 for the Fig. 3. Plot of d C values for CH4 versus CH4 /C2 ratios adapted 13 other samples), a mixing line through the remainder of from Hunt (1996). Data fall along mixing lines between the most C- the data to a 13C-enriched end-member with the d13C enriched CH4 abiogenic end-member at each site as suggested by Fig. CH4 13 + 1B, and a C-depleted CH4-rich microbial end-member, with a value value of CCS4546 and a CH4 /C2 ratio of 6 indicates that x of 55 (see text). Mixing lines are calculated based on the a mixture of up to 23% microbial CH4 can account for the equations: rest of the data at that site. 13 13 13 13 d C ¼ xðd CmicrobialÞþð1 xÞðd CabiogenicÞ; The most C-enriched end-members at each site are + too C2 -rich to be microbial in origin (Fig. 3), and where x is the percent of the microbial end-member; and, 2 likewise too H-depleted in CH4 to be consistent with 1 CH4 =C2þ¼ðxðImicrobialÞþð1 xÞðIabiogenicÞÞ ; a thermogenic origin (Fig. 2). Analogous to the Kidd Creek gases, the carbon and hydrogen isotopic patterns + where I =C2 +/CH4, the inverse of CH4 /C2. The estimated percent of these gases (Fig. 1B) suggest an abiogenic origin. microbial component for each sample (see text) is calculated by pro- 13 The fact that the data from these sites in both Canada jecting its d C value onto the mixing lines on Fig. 3. As noted in the and South Africa fall on mixing lines in both Figs. 2 text, the estimates of % microbial gas are relatively insensitive to CH4 / + 13 C2 ratios but are dependent on the selected d C values for the two end- and 3 provides excellent support for interpretation of members, and hence provide only a conservative estimate of the these data as two-component mixing between microbial relative fraction of microbial gas. In particular, the mixing lines as depicted assume that the most 13C-enriched end-member at each site is entirely free of a microbial contribution. If the most 13C-enriched samples are already themselves the product of some mixing with 13 microbial CH4 with an even more C-enriched end-member, then the estimates of the fraction of microbial contribution would be even larger. Symbols are as in Fig. 1. MM5 and MR1 (circles) are microbial

CH4 end-members previously identified for the Witwatersrand Basin based on CH4 isotopic values, 16S rRNA gene and enrichment cultures (Ward et al., 2004). Unfortunately air contamination of sample DR9IPC + + means that C2 values are diluted to below detection limit, so a CH4 /C2 ratio cannot be calculated for this sample. Based on the similarity in isotopic composition of this sample to MR1 and MM5, it has been plotted (in brackets) close to those samples for comparison purposes. 13 + Error bars are F0.5x on d C values and F7x on CH4 /C2 ratios and are smaller than the plotted symbols. lines could be recalculated for a microbial end-member with a d13C value close to MM5 (60x)(Fig. 4) CH4 13 + rather than close to DR9IPC (55x), and the data are Fig. 4. Plot of d C values for CH4 versus CH4 /C2 ratios adapted from Hunt (1996). In this plot, mixing lines are calculated as in Fig. 3 still consistent with this simple two-component mixing 13 but for mixing between the most C-enriched CH4 end-member at 13 between microbial and abiogenic gas, although the esti- each site, and a C-depleted CH4 rich microbial end-member with a mates of the fraction of microbial mixing vary some- value of 60x (see text). 336 B. Sherwood Lollar et al. / Chemical Geology 226 (2006) 328–339 and abiogenic end-members. Alternatively the presence years, re-sampling of boreholes in Canada has shown of a thermogenic component at the sites in this study that the high H2 levels are not simply an artefact of cannot be completely ruled out. Indeed, Ward et al. drilling (Bjerg et al., 1997) but a persistent natural (2004) showed that mixing between microbial and phenomenon in this geologic environment (Sherwood thermogenic gases might account for gases in boreholes et al., 1988; Sherwood Lollar et al., 1993b, 2002). Like at Evander mine located in the quartzite, shale and the hydrocarbon gases, the high quantities of H2 are minor volcanics of the Witwatersrand Supergroup sedi- likely the product of abiogenic water–rock interactions ments. Samples from South Africa in the present study in these geologically isolated, high rock to water ratio however are predominantly from boreholes in a differ- environments. Specific H2-generating reactions may ent geologic unit, the basaltic lavas of the Ventersdorp vary from one geological environment to another. At Supergroup. Furthermore, all the samples at Evander sites such as Kidd Creek where significant sections of fall close to the conventional microbial and thermo- ultramafic rock occur, H2 may be the product of ser- genic isotope fields in Fig. 2 (Ward et al., 2004), unlike pentinization over geologically long time scales analo- the 13C-enriched and 2H-depleted end-members at the gous to systems described elsewhere (Abrajano et al., sites in the present study. Hence we feel the best 1988; Charlou and Donval, 1993; Kelley et al., 2001). explanation for the data in the present study is indeed In the absence of significant ultramafics, such as at the two-component mixing between 13C-enriched abiogen- Witwatersrand Basin sites, radiolytic decomposition of ic and 13C-depleted microbial end-members. The fact water in the uranium-rich Witwatersrand Formation has that data from this study fit a simple two-component been shown to produce H2 at a sufficient rate to account mixing model between these microbial and abiogenic for the elevated H2 concentrations observed in these end-members suggests that if a thermogenic component isolated fracture fluids (Lin et al., 2005). H2 as an is present, it is small and does not control the isotopic energy source for autotrophic microbial ecosystems and compositional variations in a major way. on both the Earth (Anderson et al., 1998; Stevens and McKinley, 1995) and Mars (Boston et al., 1992) has 4.3. H2-based autotrophy linked to methanogenesis been a topic of considerable debate in the literature. Nonetheless, the number of studies that have been able This approach to resolving mixing between abiogen- to convincingly demonstrate coupled oxidation of H2 ic and microbial CH4 in the crust has important impli- and reduction of CO2 to produce CH4 via the equation: cations for evolving research into the Earth’s deep biosphere, specifically the identification of potential 4H2 þ CO2 YCH4 þ 2H2O ð1Þ substrates supporting microbial communities in the deep subsurface. H2-based subsurface microbial com- as the basis for a microbial food chain in the deep munities associated with autotrophic methanogenesis terrestrial subsurface has been limited (Chapelle et al., have been proposed in the Columbia River basalt aqui- 2002). fer (Anderson et al., 1998; Stevens and McKinley, The approach for identifying abiogenic and microbial 1995), Lidy Hot springs (Chapelle et al., 2002); and CH4 end-members presented in this paper, and the large in carbonate-rich serpentinized peridotites at an off-axis database on deep subsurface gases developed by our hydrothermal vent field near the mid-Atlantic ridge research in Canada and South Africa provides an oppor- (Kelley et al., 2001; 2005). Interest in such systems is tunity to test this proposed reaction on a large scale. high since they may be modern terrestrial analogues for Chemolithoautotrophs growing on H2 can potentially subsurface microbial ecosystems under reducing condi- reduce a variety of different oxidants including any tions on the early Earth (Shock and Schulte, 1998)or available oxygen, sulfur and nitrogen oxidants, metals on other planets or Jovian satellites (Chapelle et al., and metalloids, as well as CO2.IfH2 autotrophy linked to 2002; McCollom, 1999). Several of the boreholes sam- CO2 reduction and methanogenesis is an important pro- pled in this study have high concentrations of H2 gas as cess, there may be a correlation between H2 utilization 13 well as the CH4 and higher hydrocarbons already dis- and production of an isotopically C-depleted microbial cussed. H2 concentrations in samples from Mponeng, CH4. Fig. 5 is a compilation of all samples for Canada Driefontein and Kloof are as high as 11.5% (Table 1). and South Africa measured by our group to date (n =65), H2 levels in samples from CCS, Kidd Creek and other incorporating data from almost two dozen sites. A rela- sites in the Precambrian Shield rocks of Canada have tionship exists between H2 concentrations and the isoto- been reported in concentrations as high as 1–58% pically distinct microbial and abiogenic hydrocarbon gas (Table 1;(Sherwood Lollar et al., 1993b)). Over several end-members identified in this study. With few excep- B. Sherwood Lollar et al. / Chemical Geology 226 (2006) 328–339 337

5. Conclusions

This study suggests that the abiogenic hydrocarbon gases first identified at Kidd Creek in fact exist at a variety of Precambrian Shield sites worldwide. Using the mixing models outlined here, the relative contribution of abiogenic versus microbial end-members can be estimated. Verification of an abiogenic component first requires identification of the end-member based on the d13C and d2H model illustrated in Fig. 1, followed by evaluation of the extent of mixing with other more conventional hydrocarbon sources via mixing schemes such as in Figs. 2 and 3. This combined approach has important implications for evaluation of potential abiogenic hydrocarbon reservoirs in a wide range of geologic settings, including the longstanding controversy 13 concerning the possible contribution of abiogenic gases Fig. 5. Histogram of H2 concentrations versus d CCH4 values for samples (n =65) from this study as well as for previously reported to economic petroleum hydrocarbon reservoirs (Gold, data from Precambrian Shield sites in South Africa (Ward et al., 2004) 1979; Kenney et al., 2002). The association of high and Canada (Montgomery, 1994; Sherwood Lollar et al., 1993a,b, 13 concentrations of H2 with C-enriched CH4 end-mem- 2002). With few exceptions, high H2 concentrations (N1%) (n =31) 13 13 bers, and the H2 depletion in C-depleted methano- are associated with the most C-enriched CH4 end-members 13 x x genic end-members further suggests the possibility that (d CCH4 values between 45 and 25 ). In contrast, samples 13 with C-depleted CH4 values (indicative of a significant methano- abiogenic gases may support H2 autotrophy linked to genic component) have H2 values between 0.1–1% (n =7), to below methanogenesis in the deep subsurface. detection limit (b0.01%) (n =27). Acknowledgements tions, high H2 concentrations (N1%) are associated with This study was supported in part by grants from the 13 13 only the most C-enriched CH4 samples (d C values Natural Sciences and Engineering Research Council of between 45 and 25x). In contrast, samples with Canada and Discovery Grants Program, and Canadian 13 C-depleted CH4 values (indicative of a significant Space Agency to BSL and by the National Science methanogenic component) correlate with H2 values in Foundation Life in Extreme Environments Program the range of 0.1% to 1%, to below detection limit (EAR-9714214) grant to TCO and NASA Astrobiology (b0.01%). The observed pattern of H2 concentrations Institute grant to the IPTAI team. We thank H. Li for and CH4 isotopic signatures may be the geochemical compositional and isotopic analyses. Special thanks are fingerprint of H2 autotrophy linked to methanogenesis due to the geologists and staff of the following mines for via Eq. (1). This evidence supports the interpretation of providing geological information and invaluable assis- the hydrocarbon isotopic data as controlled by two- tance with underground field work: Kidd Creek Mine, 13 component mixing between C-enriched CH4 end- Timmins, Ontario (D. Duff, R. Cook, P. Olsen); Copper members and 13C-depleted microbial end-members Cliff South Mine, Sudbury, Ontario (T. Little); and in and may further suggest that methanogenesis is sup- South Africa, Driefontein (D. Nel, B. Kotze, M. deKo- ported by H2 utilization. A scenario could be envisaged ker), Mponeng (D. Kirshaw, H. Cole, G. Gilchrest) and where H2–CH4 rich abiogenic gases are mobilized Kloof (A. Van Heerden, D. Steinkamp, T. Hewitt, G. when fractures open due to tectonic activity or due to Buxton) Mines. Special thanks are owed to Rob Wilson more recent disturbances due to mining activity and of SRK-Turgis for logistical support. [SG] support a redox gradient-driven ecosystem in the Pre- cambrian Shield deep subsurface where rapid utiliza- References tion of H2 by microbial communities quickly depletes H and produces a 13C-depleted CH signature charac- 2 4 Abrajano, T.A., Sturchio, N.C., Bohlke, J.K., Lyon, G.L., Poreda, teristic of methanogenesis that mixes with and partially 13 R.J., Stevens, C.M., 1988. Methane–hydrogen gas seeps, Zam- overprints the pre-existing C-enriched abiogenic CH4 bales Ophiolite, Philippines: deep or shallow origin? Chemical signature. Geology 71, 211–222. 338 B. Sherwood Lollar et al. / Chemical Geology 226 (2006) 328–339

Altschul, S.F., Madden, T.L., Schaffer, A.A., Zhang, J., Zhang, Z., Hales, B.A., Edwards, C., Ritchie, D.A., Hall, G., Pickup, R.W., Miller, W., Lipman, D.J., 1997. Gapped BLAST and PSA- Saunders, J.R., 1996. Isolation and identification of methano- BLAST: a new generation of protein database search program. gen-specific DNA from blanket bog peat by PCR amplification Nucleic Acids Research 25, 3389–3402. and sequence analysis. Applied and Environmental Microbiology Anderson, R.B., 1984. The Fischer–Tropsch Synthesis. Academic 62, 668–675.

Press. Holloway, J.R., 1984. Graphite–CH4–H2O–CO2 equilibria at low Anderson, R.T., Chapelle, F.H., Lovley, D.R., 1998. Evidence against grade metamorphic conditions. Geology 12, 455–458. hydrogen-based microbial ecosystems in basalt aquifers. Science Horita, J., Berndt, M.E., 1999. Abiogenic methane formation and 281 (5379), 976–977. isotopic fractionation under hydrothermal conditions. Science Baker, B.J., Moser, D.P., MacGregor, B.J., Fishbain, S., Wagner, M., 285, 1055–1057. Fry, N.K., Jackson, B., Speolstra, N., Loos, S., Takai, K., Sher- Hu, G., Ouyang, Z., Wang, X., Wen, Q., 1998. Carbon isotope wood Lollar, B., Fredrickson, J.K., Balkwill, D., Onstott, T.C., fractionation in the process of Fischer–Tropsch reaction in prim- Wimpee, C.F., Stahl, D.A., 2003. Related assemblages of sul- itive solar nebula. Science China 41, 202–207. phate-reducing bacteria associated with ultradeep gold mines of Hunt, J.M., 1996. Petroleum Geochemistry and Geology. W.H. Free- South Africa and deep basalt aquifers of Washington State. En- man and Company, New York. 743 pp. vironmental Microbiology 5 (4), 267–277. Jenden, P.D., Kaplan, I.R., 1989. Origin of natural gas in the Sacra-

Berndt, M.E., Allen, D.E., Seyfried, W.E.J., 1996. Reduction of CO2 mento Basin, California. American Association of Petroleum during serpentinization of olivine at 300 8C and 500 bar. Geology Geologists Bulletin 73, 431–453. 24 (4), 351–354. Jenden, P.D., Kaplan, I.R., Poreda, R.J., Craig, H., 1988. Origin of Bjerg, P.L., Jakobsen, R., Bay, H., Rasmussen, M., Albrechtsen, H.-J., nitrogen-rich natural gases in the California Great Valley: evi- Christensen, T.H., 1997. Effects of sampling well construction on dence from helium, carbon and nitrogen isotope ratios. Geochi-

H2 measurements made for characterization of redox conditions in mica et Cosmochimica Acta 52, 851–861. a contaminated aquifer. Environmental Science and Technology Jenden, P.D., Hilton, D.R., Kaplan, I.R., Craig, H., 1993. Abiogenic 31, 3029–3031. hydrocarbons and mantle helium in oil and gas fields. In: Howell, Boston, P.M., Ivanov, M.V., McKay, C.P., 1992. On the possibility of D.G. (Ed.), The Future of Energy Gases, U.S. Geological Survey chemosynthetic ecosystems in subsurface habitats on Mars. Icarus Professional Paper, vol. 1570, pp. 31–56. 95, 300–308. Kelley, D.S., Karson, J.A., Blackman, D.K., Fruh-Green, G.L., Chapelle, F.H., O’Neill, K., Bradley, P.M., Methe, B.A., Ciufo, S.A., Butterfield, D.A., Lilley, M.D., Olson, E.J., Schrenk, M.O., Knobel, L.L., Lovley, D.R., 2002. A hydrogen-based subsurface Roe, K.K., Lebon, G.T., Rivizzigno, P., 2001. An off-axis hy- microbial community dominated by methanogens. Nature 415, drothermal vent field near the Mid-Atlantic Ridge at 308N. 312–315. Nature 412, 145–149. Charlou, J.L., Donval, J.P., 1993. Hydrothermal methane venting Kelley, D.S., Karson, J.A., Fruh-Green, G.L., Yoerger, D.R., Shank, between 12N and 26N along the Mid-Atlantic ridge. Journal of T.M., Butterfield, D.A., Hayes, J.M., Schrenk, M.O., Olson, E.J., Geophysical Research 98, 9625–9642. Proskurowski, G., Jakuba, M., Bradley, A., Larson, B., Ludwig, Cochrane, L.B., 1984. Ore deposits of the Copper Cliff offset. In: K., Glickson, D., Buckman, K., Bradley, A.S., Brazelton, W.J., Kendrick, G. (Ed.), The Geology and Ore deposits of the Sudbury Roe, K., Elend, M.J., Delacour, A., Bernasconi, S.M., Lilley, Structure. Ontario Geological Survey, pp. 347–360. M.D., Baross, J.A., Summons, R.T., Sylva, S.P., 2005. A serpen- Cook, A.P., 1998. The occurrence, emission and ignition of com- tinite-hosted ecosystem: the Lost City Hydrothermal Field. Sci- bustible strata gases in Witwatersrand gold mines and Bushveld ence 307, 1428–1434. platinum mines and means of ameliorating related ignition and Kenney, J.F., Kutcherov, V.A., Bendeliani, N.A., Alekseev, V.A., explosion hazards. Safety in Mines Research Advisory Com- 2002. The evolution of multicomponent systems at high pressures: mittee, Interim Project Report, vol. 504. Itasca Africa (Pty) Ltd. the thermodynamic stability of the hydrogen-carbon system: the 90 pgs. genesis of hydrocarbons and the origin of petroleum. PNAS 99 Coward, M.P., Spencer, R.M., Spencer, C.E., 1995. Development of (17), 10976–10981. the Witwatersrand Basin, South Africa. Early Precambrian Pro- Kerr, R.A., 2004. Heavy breathing on Mars? Science 306, 29. cesses 95, 243–269. Kieft, T.L., Fredrickson, J.K., Onstott, T.C., Gorby, Y.A., Kostandar- Des Marais, D.J., Donchin, J.H., Nehring, N.L., Truesdell, A.H., ithes, H.M., Bailey, T.J., Kennedy, D.W., Li, S.W., Plymale, A.E., 1981. Molecular carbon isotopic evidence for the origin of geo- Spadoni, C.M., Gray, M.S., 1999. Dissimilatory reduction of thermal hydrocarbons. Nature 292, 826–828. Fe(III) and other electron acceptors by a Thermus isolate. Applied Doig, F., 1994. Bacterial methanogenesis in Canadian Shield ground- and Environmental Microbiology 65 (3), 1214–1221. waters. M.Sc. Thesis, University of Toronto, Toronto. 99 pp. Kieft, T.L., McCuddy, S.M., Onstott, T.C., Davidson, M., Lin, L.-H., Formisano, V., Atreva, S., Encrenaz, T., Ignatiev, N., Giuranna, M., Mislowack, B., Pratt, L., Boice, E., Sherwood Lollar, B., Lipp- 2004. Detection of methane in the atmosphere of Mars. Science mann-Pipke, J., Pfiffner, S.M., Phelps, T.J., Gihring, B., Moser, 306, 1758–1761. D., van Heerden, A., 2005. Geochemically generated, energy-rich Foustoukos, D.I., Seyfried, W.E.J., 2004. Hydrocarbons in hydro- substrates and indigenous microorganisms in deep, ancient thermal vent fluids: the role of chrome-bearing catalysts. Science groundwater. Geomicrobiology Journal 22, 325–335. 304, 1002. Lahermo, P.W., Lampen, P.H., 1987. Brackish and saline ground- Giardini, A.A., Salotti, C.A., 1969. Kinetics and relations in the calcite– waters in Finland. In: Fritz, P., Frape, S.K. (Eds.), Saline Water hydrogen reaction and relations in the dolomite–hydrogen and sid- and Gases in Crystalline Rock, Geological Association of Canada erite–hydrogen systems. American Mineralogist 54, 1151–1172. Special Paper, vol. 33, pp. 103–110. Gold, T., 1979. Terrestrial sources of carbon and earthquake out- Lin, L.-H., Slater, G.F., Sherwood Lollar, B., Lacrampe-Couloume, gassing. Journal of Petroleum Geology 1, 1–19. G., Onstott, T.C., 2005. The yield and isotopic composition of B. Sherwood Lollar et al. / Chemical Geology 226 (2006) 328–339 339

radiolytic H2, a potential energy source for the deep subsurface Rigby, D., Smith, J.W., 1981. An isotopic study of gases and hydro- biosphere. Geochimica et Cosmochimica Acta 69, 893–903. carbons in the Cooper Basin. Australian Petroleum Exploration Lippmann, J., Stute, M., Torgersen, T., Moser, D.P., Hall, J., Lin, L., Association Journal 21, 222–229. Borcsik, M., Bellamy, R.E.S., Onstott, T.C., 2003. Dating ultra- Schoell, M., 1988. Multiple origins of methane in the Earth. Chemical deep mine waters with noble gases and 36Cl, Witwatersrand Basin, Geology 71, 1–10. South Africa. Geochimica et Cosmochimica Acta 67, 4597–4619. Sherwood, B., Fritz, P., Frape, S.K., Macko, S.A., Weise, S.M., Lyon, G.L., Hulston, J.R., 1984. Carbon and hydrogen isotopic Welhan, J.A., 1988. Methane occurrences in the Canadian Shield. compositions of New Zealand geothermal gases. Geochimica et Chemical Geology 71, 223–236. Cosmochimica Acta 48, 1161–1171. Sherwood Lollar, B., Frape, S.K., Fritz, P., Macko, S.A., Welhan, McCollom, T.M., 1999. Methanogenesis as a potential source of J.A., Blomqvist, R., Lahermo, P.W., 1993a. Evidence for bacteri- chemical energy for primary biomass production by autotrophic ally generated hydrocarbon gas in Canadian Shield and Fennos- organisms in hydrothermal systems on Europa. Journal of Geo- candian Shield rocks. Geochimica et Cosmochimica Acta 57, physical Research 104 (E12), 30729–30742. 5073–5085. McCollom, T.M., Seewald, J.S., 2001. A reassessment of the potential Sherwood Lollar, B., Frape, S.K., Weise, S.M., Fritz, P., Macko, S.A.,

for reduction of dissolved CO2 to hydrocarbons during serpenti- Welhan, J.A., 1993b. Abiogenic methanogenesis in crystalline nization of olivine. Geochimica et Cosmochimica Acta 65 (21), rocks. Geochimica et Cosmochimica Acta 57, 5087–5097. 3769–3778. Sherwood Lollar, B., Frape, S.K., Weise, S.M., 1994. New sampling

Montgomery, J., 1994. An isotopic study of CH4 and associated N2 devices for environmental characterization of groundwater and and H2 gases in Canadian Shield mining environments. MSc dissolved gas chemistry (CH4,N2, He). Environmental Science Thesis, University of Toronto, Toronto, Canada, 72 pp. and Technology 28 (13), 2423–2427. Moser, D.P., Gihring, T.M., Brockman, F.J., Fredrickson, J.K., Balk- Sherwood Lollar, B., Westgate, T.D., Ward, J.A., Slater, G.F., will, D.L., Dollhopf, M.E., Sherwood Lollar, B., Pratt, L.M., Lacrampe-Couloume, G., 2002. Abiogenic formation of gaseous Boice, E., Southam, G., Wander, G., Baker, B.J., Pfiffner, S.M., alkanes in the Earth’s crust as a minor source of global hydrocar- Lin, L.-H., Onstott, T.C., 2005. Deep continental fracture system bon reservoirs. Nature 416, 522–524. dominated by Methanobacterium spp. and Desulfotomaculum Shock, E.L., 1995. An open or shut case? Nature 378, 338–339. spp. Applied and Environmental Microbiology. In press. Shock, E.L., Schulte, M.D., 1998. Organic synthesis during fluid Moyer, C.L., Tiedje, J.M., Dobbs, F.C., Karl, D.M., 1998. Diversity mixing in hydrothermal systems. Journal of Geophysical Research of deep-sea hydrothermal vent Arachaea from Loihi Seamount, 103 (E12), 28513–28527. Hawaii. Deep-Sea Research II 45, 303–317. Stevens, T.O., McKinley, J.P., 1995. Lithoautotrophic microbial eco- Nurmi, P.A., Kukkonen, I.T., 1986. Geochemistry of water and gas systems in deep basalt aquifers. Science 270, 450–454. from deep drill holes: a new sampling technique. Canadian Jour- Takai, K., Moser, D.P., DeFlaun, M., Onstott, T.C., Fredrickson, J.K., nal of Earth Sciences 23, 1450–1454. 2001. Archaeal diversity in waters from deep South African Onstott, T.C., Moser, D.P., Pfiffner, S.M., Fredrickson, J.K., Brock- gold mines. Applied and Environmental Microbiology 67 (12), man, F.J., Phelps, T.J., White, D.C., Peacock, A., Balkwill, D., 5750–5760. Hoover, R., Krumholz, L.R., Borscik, M., Kieft, T.L., Wilson, R., Vanko, D.A., Stakes, D.S., 1991. Fluids in oceanic layer 3: evidence 2003. Indigenous and contaminant microbes in ultradeep mines. from veined rocks, hole 735B, Southwest Indian Ridge. Proceed- Environmental Microbiology 5, 1168–1191. ings of the Oceanic Drilling Program. Scientific Results 118, Oremland, R.S., Des Marais, D.J., 1983. Distribution, abundance and 181–215. carbon isotopic composition of gaseous hydrocarbons in Big Soda Ward, J.A., Slater, G.F., Moser, D.P., Lin, L.-H., Lacrampe-Cou- Lake, Nevada: an alkaline, meromictic lake. Geochimica et Cos- loume, G., Bonin, A.S., Davidson, M., Hall, J.A., Mislowack, mochimica Acta 47, 2107–2114. B.J., Bellamy, R.E.S., Onstott, T.C., Sherwood Lollar, B., 2004. Poreda, R.J., Jenden, P.D., Kaplan, I.R., Craig, H., 1986. Mantle Microbial hydrocarbon gases in the Witwatersrand Basin, South helium in Sacramento basin natural gas wells. Geochimica et Africa: implications for the deep biosphere. Geochimica et Cos- Cosmochimica Acta 50, 2847–2853. mochimica Acta 68 (13), 3239–3250. Poreda, R.J., Jeffrey, A.W.A., Kaplan, I.R., Craig, H., 1988. Mag- Yuen, G.U., Pecore, J.A., Kerridge, J.F., Pinnavaia, T.J., Rightor, matic helium in subduction-zone natural gases. Chemical Geology E.G., Flores, J., Wedeking, K., Mariner, R., Des Marais, D.J., 71, 199–210. Chang, S., 1990. Carbon isotope fractionation in Fischer– Reysenbach, A.L., Longnecker, K., Kirshtein, J., 2000. Novel bacte- Tropsch type reactions. Lunar Planetary Science Conference rial and archaeal lineages from an in situ growth chamber XXI, pp. 1367–1368. deployed at a Mid-Atlantic Ridge hydrothermal vent. Applied and Environmental Microbiology 66, 3798–3806.